In the first place, if you had read the linked material, you would note that I published a “costing-out model.” This is something more than a mere opinion piece. The essential point is that as you move away from the customer, towards the center of the network, economies of scale cut in. Once you get to the point where a given piece of equipment serves a hundred, or a thousand, or a million people simultaneously, their pro-rata share of its cost is not very great. The cost of this piece of equipment becomes small, compared to the total cost of the many subscriber loops it serves. If you take something like an AT&T U-Verse cabinet, serving a hundred houses, then thousands of dollars of installation cost per cabinet work out to tens of dollars per subscriber, amortized at a rate of quarter-dollars per month. Of course, the vast majority of the network, everything downstream from the cabinet, gets left alone. As for the back-haul, according to my calculation, each subscriber’s pro-rata share of the optical fiber leading from the U-Verse cabinet to the rest of the world is on the order of twenty feet, ten feet of it between the cabinet and the central office, and ten feet of it between the central office and the world. he former portion will cost about a hundred dollars (five dollars a year, fifty cents a month a month). Much of that twenty feet, especially the upstream part is already in the form of pre-existing ducts (see below).

Please go and read Jeff Hecht, Understanding Fiber Optics (1987). In particular, read the discussion of graded-index, single-mode fiber, which has long since become the industry standard. The theoretical capacity of an optical fiber is something like 600 T-bits (600 million megabits), and the Japanese have actually achieved 30 T-bits at last report, over a distance of a hundred miles. New modems, transmitting and receiving on a greater number of frequencies, can massively increase the capacity of a fiber which is already in the ground. It is not a question of “adding more trunk line.” A practical infinity of trunk line is already there. The conditions under which additional optical fibers would be needed are extremely rare, and are not commonly found in the “last mile” An optical fiber in the “last mile” is, almost by definition, grossly under-utilized.

In any case, trunk cables are installed inside plastic ducts. A trench is dug, one or more ducts are placed inside it, and cables are subsequently blown through the ducts by compressed air, or pulled by messenger strings (), without having to dig the ducts up again. Feeding the cable through the duct is many times cheaper than digging the trench, etc. For that matter, methods have been developed to pump pressurized lubricants into existing copper cables in the ground, which may not have been installed in ducts, and push the interior components out, leaving the jacket in place as a duct suitable for an optical cable.

() I, personally, have watched the latter operation, over a distance of fifty yards, from a telephone cabinet to a point where the cable was subsequently carried on existing overhead electrical poles, in order to provide service to a new office building in what had previously been a low-density residential district (bungalows on acre-lots). The men just pulled the cord, hand over hand, without mechanical aids, and it took a very few minutes.

In my model, I was referring primarily to telephone-type networks. The situation is a bit different in cable television. What one would do is to cease broadcasting a hundred channels of unsolicited content over the shared local loop, leading to fewer than a hundred houses, and transmit only information which customers had specifically asked for, on a second-by-second basis. One might think of it as the difference between a newspaper boy throwing a newspaper on every porch, and a postman placing particularly addressed letters, magazines, parcels, and newspapers in particular mailboxes. Upstream from the street corner, of course, the situation in cable becomes substantially identical to that in telephony, as economies of scale cut in.